Research Seminar
Model Selection and Inference in Copula IRT modelsJohan BraekenKU Leuven | |
| Abstract: | In recent work copula IRT models were developed to accommodate for Local Item Dependencies (e.g., due to a contextualized test format), which are violations of local stochastic independence, a key assumption underlying most measurement models. In this seminar session, we will firstly discuss how to select between various copula IRT models. This is not trivial due to the models being non-nested, having an equal amount of parameters, yet a different functional form. Therefore, several model selection approaches (e.g., Minimum Description Length and model weights within a continuous model expansion) will be explored. Secondly, the supplementary question is how robust the IRT model is to misspecification of the LID structure. A simulation study to evaluate these two questions will be reported and might possibly support a quite bold statement about modeling LID. |
| Date: | Tue Oct 30, 12:15 pm - 1:15 pm |
| Place: | room 00.60 (Department of Psychology, Tiensestraat 102, 3000 Leuven) |